MATLAB Answers

0

Low pass filter design first peak elimination

SAMET YILDIZ さんによって質問されました 2014 年 9 月 6 日
Hello Everyone, I try to make a low pass operation on my data set which has 1kHz sampling rate. When I try to make fft power spectrum, I always see the first dominant peak and cannot observe the other details clearly. My advisor wants from me to use a low pass filter to eliminate this highly dominant peak of the power spectrum. I used fdatool with Fp = 800, Fs = 1600, Fsampling = 10000, Apass = 1dB, Astop = 80dB. The filter has the freq response as shown in Fig 1.
*Fig 1*
As I apply this filter to my data set, which gives the results like below in Fig 2. The green line is the original data set and the blue which has another unwanted occured peak one is the filtered data. This peak is not related with anything. It came out after filtering. Simply, I need to get rid of this with a proper filter design. Can you help me with this please ?
Fig 2
Thanks

  3 件のコメント

Please post your MATLAB code.
One item I noticed earlier: ‘... data set which has 1kHz sampling rate ...’ then later ‘Fsampling = 10000’.
Either there is a typographical error in your post, you omitted some information (like upsampling), or the filter your are designing is not the one you believe you are designing.
Hello, simply I have 65000 sample data set which has been gotten by 1kHz sampling rate. The picture below represents it.
As I make the fft power spectrum in db scale, I get the result like below.
According to what I learnt from my advisor, if I can apply a good low pass filter, I can get rid of this really high peak value of fft spectrum in low frequencies. That is why, I can have better picture as can be seen that there is a small waving effect in the fft spectrum, they would like to see this effect. I hope you understand what I try to tell.
The code that I used for the fft is below:
_ _function y = fftGraph( file, fs )
% this program gets the text file which is written 'text.extension'
% and the number of the column which is wanted to process.
% After reading the column, the program plots the real time plot,
% power spectrum in frequecny domain and logarithmic plot of the
% power spectrum.
y = file;
N=numel(y);
ts=1/fs;
tmax=(N-1)*ts;
t=0:ts:tmax;
f=-fs/2:fs/(N-1):fs/2;
z=fftshift(fft(y));
re=real(z);
im=imag(z);
msa=2*(re.^2+im.^2)/(N^2);
loglog(f,msa,'Marker','.','Color','k');_ _
Thanks

サインイン to comment.

1 件の回答


your question is very vague, need more plots and details.

  0 件のコメント

サインイン to comment.



Translated by